Abstract
Theory suggests that the interannual equatorial Kelvin wave sea level (ηW) in the western Pacific should be directly proportional to the interannual western Australian sea level, and the interannual equatorial Kelvin wave sea level (ηE) in the eastern Pacific should be directly proportional to the equatorial interannual sea level at the South American coast. Forced equatorial Kelvin wave dynamics implies, approximately, that
EttWtBtt
where Δt, τ¯, B, and θ, are, respectively, a lag corresponding to equatorial Kelvin wave propagation, zonally averaged equatorial interannual eastward wind stress, a forcing coefficient and a number between 0 and 1. A regression analysis of the sea level difference above using western Australian and South American sea levels against τ¯ shows that the correlation is high and is highest (0.83) when &Deltat=3 months and θ&Deltat=1 month. The values for &Deltat, θ, and B are reasonable for they are appropriate for a signal dominated by the first and second vertical modes and forced mainly slightly west of the central equatorial Pacific.
Other relationships between ηW, ηE, and τ¯ are examined in addition to (1) to test some common ENSO ideas. An important component of the delayed oscillator theory of ENSO is that reflected Rossby waves at the western Pacific Ocean boundary should trigger wind changes in the equatorial Pacific, that is, ηW should lead τ¯. Surprisingly, the authors’ correlation analysis shows that ηW is not significantly positively correlated with τ¯ at any lag or lead. It seems that while reflection of Rossby waves at the western boundary may influence ocean and atmosphere dynamics in the Pacific interior, extra physics beyond delayed oscillator theory is necessary to describe ENSO. However, we found that other ideas about ENSO were consistent with the data. In agreement with the idea that strengthening trade winds result in a build up of sea level in the western equatorial Pacific and also in agreement with Rossby wave and equatorial wave reflection dynamics, ηW is maximally correlated with −τ¯ when it lags −τ¯ by the theoretically reasonable value of about 3 months. In accordance with the notion that El Niño results when there is a relaxation of the trade winds, ηE is highly correlated with τ¯. But analysis also shows that the commonly held idea that ηE should lag τ¯ because of the equatorial Kelvin wave propagation from the main region of forcing in the west-central equatorial Pacific is questionable because of the influence of the out of phase reflected equatorial Kelvin wave signal ηW.
